JP3205220B2 - Data retrieval device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data search device used for searching desired data from a search target database in which data is stored.

[0002]

2. Description of the Related Art A data retrieval apparatus retrieves desired data from a retrieval target database in which a large number of data are stored, based on a retrieval request item. Each data stored in the search target database is usually set with a relative similarity, and when a search request item is input by an input means such as a keyboard, each search request item and each of the data stored in the database are input. The degree of similarity with the data is determined based on each similarity. Then, data having a high degree of similarity to the search request item is extracted from the search target database and output to a display device such as a display.

The input means for inputting the search request item is generally a text input device such as a keyboard. However, the input means is directly input by image, voice, or the like, or the attribute of the search request item is set by a plurality of numerical values. There is also a method of inputting.

Generally, a plurality of keywords are set as the relative similarity of each data stored in the search target database. In this case, the degree of similarity between the keyword of the search request item and the keyword in each data is determined based on the degree of matching. There is also known a method in which each data and search request item in a database is represented by a multidimensional vector by quantifying the similarity for each of a plurality of elements. In this case, the degree of similarity is determined based on the Euclidean distance, Mahalanobis distance, and the like between the vector of the search request item and each data and vector.

[0005] The search result output means is not limited to a method of printing out the corresponding data, but also a method of outputting the data by ranking or scoring based on the degree of similarity. Also,
As disclosed in JP-A-1-209528, a method of visually expressing the retrieved data in a two-dimensional or three-dimensional space by coordinate transformation such as principal component analysis is also employed.

[0006]

As described above, each data stored in the search target database is one-dimensionally searched based on the degree of similarity with the search request item. For this reason, even if there is a plurality of data items that are determined to have the same degree of similarity to the search request item, those data items are not necessarily required to the same degree as the search result for the search request item, Therefore, even unnecessary data is searched.

[0007] For example, as an item to be searched, "animal"
Is input, "lion" and "dolphin" as data have similar degrees of similarity from the viewpoint of "animals". However, "lions" are terrestrial animals, while "dolphins" are aquatic animals. From such a viewpoint, the degree of similarity cannot be said to be high. Therefore, when a terrestrial animal is requested as a search result, simply inputting "animal" as a search request item will output "lion" and "dolphin" with a similar degree of similarity. . As a result, the output search results include unnecessary data, and there is a problem that effective search results cannot be obtained efficiently.

The present invention has been made to solve such a problem, and an object of the present invention is to provide a data search device capable of efficiently searching for required data.

[0009]

According to the present invention, there is provided a data search apparatus comprising: a search target database storing a plurality of data sets each having a relative similarity; input means for inputting a search request item; Determining the degree of similarity between each data stored in the search target database and the search request item input by the input means based on the similarity of each data and search request item, Search means for extracting a plurality of data similar to the search request item, display means for displaying the plurality of data extracted by the search means together with the search request item, and data displayed on the display means. Operation means for moving the search request item closer to or farther from the search request item. And data moved by the stage, the degree of similarity between the other data displayed on the display means,
Judgment is made based on the similarity of each data, and based on the judgment, the other data displayed on the display means is moved with respect to the search target item,
Thereby, the above object is achieved.

The movement distance of the data displayed on the display means increases as the degree of similarity to the data moved by the operation means increases.

Further, each data stored in the search target database is represented by a vector in which the similarity is digitized for each of a plurality of elements.

[0012]

In the data search device of the present invention, when a search request item is input by the input means, the similarity between the two items is determined based on the similarity of the item and the similarity of each data stored in the search target database. Is calculated, and data having a high degree of similarity to the search request item is extracted. The extracted data is displayed on the display together with the search request item. Then, from the data displayed on the display means, if the data considered necessary as a search result is moved by the operation means so as to approach the search target item, the degree of similarity to the moved data is determined. Is moved so as to approach the search target item greatly. Conversely, from the data displayed on the display means, if the data that is considered unnecessary as a search result is moved by the operation means so as to move away from the search target item,
Data having a high degree of similarity to the moved data is moved so as to be far away from the search target item. As a result, data required as a search result is displayed on the display means in a state close to the search target item.

[0013]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a block diagram showing an example of the data search apparatus of the present invention. The data search device includes an input device 1 for inputting a search request item, a search target database 2 storing a large number of data having respective relative similarities set therein, and a search request input by the input device 1. A retrieval device 3 for retrieving data from the retrieval target database 2 for an item, and a display device 4 for displaying the data retrieved by the retrieval device 3 in two-dimensional plane coordinates
And an operation device 5 for moving arbitrary data displayed in two-dimensional coordinates on the display device 4.

The search target database 2 stores many words as data to be searched. Each word stored in the search target database 2 is set by numerically expressing the relative similarity between the words for each of the six elements. Therefore, each word is represented by a six-dimensional vector. ing.

A word as a search request item (hereinafter, simply referred to as an item) is input to the input device 1. Words input as items are also stored in the search target database 2 as six-dimensional vectors.

The display device 4 is constituted by, for example, a liquid crystal display device or the like, and displays a plurality of words searched by the search device 3 together with item words in two-dimensional coordinates. In this embodiment, the display device 4
Displays a word input as an item at the origin of coordinates, and a plurality of words searched by the search device 3 are displayed around the word based on the degree of similarity to the word of the item. I have. That is, words having a high degree of similarity to the item words are displayed close to the item words, and words having a low degree of similarity to the item words are displayed away from the item words.

Each word displayed as a search result on the two-dimensional coordinates of the display device 4 is moved by operating the operation device 5 constituted by a pointing device such as a mouse and a trackball. I have. The word moved by the operation device 5 is considered to be necessary as a search result, or
Any word that is considered unnecessary as a search result is considered to be necessary as a search result. The word that is considered unnecessary as the search result is moved closer to the word of the item located at the origin. Is moved away from the word. Then, for each word displayed on the display device 4 other than the moved word, the moving distance is calculated based on the degree of similarity to the moved word, and is the word moved in the same direction as the moved word? , Is kept stopped.

FIG. 2 is a flowchart for explaining the operation of the data search device of the present invention. When a word as a search request item is input by the input device 1 (see step S1 in FIG. 2; the same applies hereinafter), the search device 3 searches for data in the search target database 2 (step S3).

The search device 3 first reads a vector indicating the relative similarity of the word of the item from the search target database 2, and reads the vector of the word of the item and each of the words stored in the search target database 2. The distance from the word vector is calculated.

In the search target database 2, each word is
It is expressed as a six-dimensional vector in which the relative similarity is quantified for each of the six elements. The vector is not limited to a numerical value of the similarity of each element, and may be a numerical value of a plurality of other attributes of a word. For example, as shown in FIG. 3, the vector V _Wx of the word W _x is represented as (v _x1, v _x2, ... _, V _x6 ) by six numerical values, and the vector V _Wy of the word W _y is , And six numerical values (v _y1, v _y2, ... _, V _y6 ). And
The degree of similarity between the words W _x and the word W _y is determined by the Euclidean distance of the vector V _Wx and vector V _Wy of each word W _x and word W _{y. Assuming} that the Euclidean distance between the vector V _Wx and the vector V _Wy is dist (V _Wx , V _Wy ), the distance dist (V _Wx , V _Wy ) is calculated based on the equation (1).

[0022]

Dist (V _Wx , V _Wy ) = ｛(v _x1 −v _y1 ) ² + (v _x2 −v _y2 ) ² + (v _x3 −v _y3 ) ² + (v _x4 −v _y4 ) ² + ^{_{(v x5 - v y5) 2}} + (v x6 - v y6) 2} vector V _I word W _I ^1/2 items, the vectors of each word W stored in the search database 2 and V _W , The search device 3 calculates a vector V _I of the word of the item,
The distance dist (V _I , V _W ) between each word W and the vector V _W is represented by
Based on the expression (1), each is calculated, and the degree of similarity between the word W _{I of} the item and each word W stored in the search target database 2 is determined.

For example, when "animal" is input as a word of an item by the input device 1, the retrieval device 3 takes out a vector (referred to as _VIa ) of the word "animal" and retrieves the vector _VIa and , The distance between each word W, which is data in the search target database 2, and the vector V _W is calculated. As shown in FIG. 3, the vector V _Ia of the word "animal" as an item is represented by (8,7,2,2,0,1).

In this case, when the word W stored in the search target database 2 is “wolf”, the vector of the word “wolf” (referred to as V _Wo ) is, as shown in FIG. has become a 8,4,1,2,0), the distance between the vector V _Ia word item "animal" is calculated based on equation (1). The arithmetic expression is shown as Expression (2).

[0025]

Dist (V _Ia , V _Wo ) = ｛(8-8) ² + (7-8) ² + (2-4) ² + (2-1) ² + (0-2) ² + ( 1-0) ² ｝ ^1/2 ≒ 3.32 (2) In this way, the search device 3 sets the vector V _W of each word W in the search target database 2 and the word W _{I of} the item.
When the distance between the vector V _I calculates each word having a word W _I vector V _I distance is relatively close vector V _W against the item, extracted several. Then, all the extracted words W are output to the display device 4 as those having a high degree of similarity to the item words (step S3). The display device 4 displays the word W _{I of} the item.
At the same time, all the extracted words W are displayed in two-dimensional coordinates. In this case, the word W _I of the item is displayed at the origin position of the two-dimensional coordinates. Further, the display position of each word W that has been extracted, a vector V _W of each word W, is determined based on the distance between the vector V _I word W _I of item, the distance between both vectors V _W and V _I dist (V _I , V _W ) is set equal to the distance from the origin to the display position of each word W.

That is, assuming that the coordinates of the display position of the word W on the two-dimensional plane of the display device 4 are (x, y), the display coordinates (x, y) of the word W satisfy the expression (3). Selected.

[0027]

(X ² + y ² ) ^1/2 = dist (V _I , V _W ) (3) The coordinates (x, y) are θ selected by random numbers.
(0 ≦ θ <2π) is determined so as to satisfy the expression (4).

[0028]

(X, y) = (dist (V _I , V _W ) cos θ, dist (V _I , V _W ) sin θ) (4) As a result, the display device 4 displays the word W of the item. _I appears at the origin, the around, along with the degree of similarity is high word W are displayed close each word W _I of item to the vector V _W word W _I of item, word items W _I vector V _W for similar degree lower word W of are respectively displayed on the far from the word W _I of item.

FIG. 4 shows an example of a display screen of the display device 4 when the word "animal" is input as an item.

In this case, the coordinates (x, y) of the display position of each word W extracted from the search target database 2 on the display device 4 are stored in a buffer. FIG. 5 shows an example of the display coordinates stored in the buffer of each displayed word W on the display screen shown in FIG. As is clear from FIG. 5, the word "wolf"
Is displayed at the position indicated by the coordinates (2.5, -2.2).

As described above, when the word W related to the word W _I of the item is extracted from the search target database 2 and displayed on the display device 4, the search result is obtained from all the displayed words W. Those that are deemed necessary and those that are deemed unnecessary as search results are selected. Then, by the operation of the operation device 5, the word W considered necessary as the search result is moved so as to approach the word W _I of the item located at the origin, and the word W considered unnecessary as the search result is It is moved away from the word W _I of the item.

When the search device 3 confirms that the word W has been moved by the operation input from the operation device 5 (step S4), the search device 3 narrows down the search data displayed on the display device 4. (Step S5). If the operation device 5 is not operated and, therefore, there is no operation input to the search device 3, it is confirmed that the search may be terminated (step S).
6), the data search operation ends.

FIG. 6 is a block diagram showing the configuration of the operation device 5 composed of a pointing device such as a mouse and a trackball. The operation device 5 displays an arbitrary word W displayed on the display device 4.
An input unit 51 operated when changing the display position of a, a calculation unit 52 that performs a predetermined calculation with reference to data stored in the search target database 2, and a calculation result by the calculation unit 52 An output unit 53 for changing a display result of the display device 4.

FIG. 7 is a flowchart showing a specific procedure for narrowing down search data by the operation device 5.

When one word W _S displayed on the display device 4 is selected and the word W _S is moved by the operation device 5, the arithmetic unit 52 of the operation device 5 causes the word W _S, together with capture a moving direction relative to the origin of the two-dimensional plane coordinates of the display device 4, and calculates the movement distance (see step S11 in FIG. 7, hereinafter the same). The moving direction with respect to the coordinate origin is positive when moving away from the coordinate origin, and negative when approaching the coordinate origin.

The moving distance of the word W _S includes the word W _S before the movement of the coordinates in the two-dimensional plane of the display device 4 is calculated by the coordinate after the movement. The coordinates before the movement of words W _S in the two-dimensional plane of the display device 4 (x ', y') , the coordinates after the movement (x ", y") When, the moving distance M _S word W _S, It is expressed by equation (5).

[0037]

M _S = ５ (x ″) ² + (y ″) ² ｝ ^1/2 − ｛(x ′) ² + (y ′) ² ｝ ^1/2 (5) In this manner, as one word W _S moves away relative to the word of the item to be located are selected coordinate origin, or if it is moved so as to approach, the arithmetic unit 52, the vector V _S of the moved words W _S When the distance between the vector V _W of each of the other word W that is displayed on the display unit 4 calculates each distance vector V _W of each word W for the vector V _S of the moved words W _S dist (V _W, so as to be inversely proportional to the V _S), and calculates the movement distance of each word W (step S12), the in the same direction as the moving direction of the word W _S with respect to the coordinate origin, moves each word W, respectively (step S
13).

The distance between the vector V _W of each word W displayed on the display device 4 and the vector V _S of the moved word W _S
dist (V _W, V _S) is small, therefore, the word W is, when the degree of similarity is high relative to the moved words W _S is ne size enough to match the travel distance of the moved words W _S It is moved by a distance in a direction away from the coordinate origin or in a direction approaching the coordinate origin. Conversely, the distance dist (V _W, V _S) of the vector V _S and the vector V _W of the moved words W _S is large, therefore, the word W is, the degree of similarity with respect to the moved words W _S If the height is low, it is moved by a small distance in a direction away from the coordinate origin or in a direction approaching the coordinate origin. In this case, the vector V _W and the moved word W
When the distance between the vector V _S of _S is larger than a predetermined value, the word W will maintain the stopped state without being moved.

Assuming that an unnecessary word W _S as a search result is moved in a direction away from the coordinate origin by operating the operation device 5, each word W displayed on the display device 4 moves only in a direction away from the coordinate origin. Or remain stationary without being moved. Also, assuming that the word W _S required as a search result is moved in the direction approaching the coordinate origin by operating the operation device 5, each word W displayed on the display device 4 moves only in the direction approaching the coordinate origin. Or remain stationary without being moved. Travel distance M _W of each word W is the distance M _S moved the moved words W _S, the distance dist between the vector V _W of the vector V _S and each word W in the word W _S (V
_W , V _S ) and is calculated by equation (6). Here, α is a coefficient (0 <α).

[0040]

When 1−α · dist (V _W , V _S )> 0, M _W = ｛1−α · dist (V _W , V _S )｝ M _S 1−α · dist (V _W , V _S ) ≦ 0 M _W = 0 (6) When the moving distance M _W of each word _W is calculated in this way, the calculation unit 52 of the operation device 5 determines that each word W has the distance M _W And the coordinates (x " _W , y" _W ) on the two-dimensional plane of the display device 4 when they are moved by
Each is calculated based on the equation (7). However, (x ' _W ,
y ′ _W ) is the coordinates of each word W before movement on the two-dimensional plane of the display device 4.

[0041]

Equation 7] _{x "W = x 'W +} M W · x' W / {(x 'W) 2 + (y' W) 2} 1/2 y" W = y 'W + M W · y' W / ｛(X ' _W ) ² + (y' _W ) ² ｝ ^1/2 ... (7) The coordinates (x ") of each word W obtained as described above after movement
_W, y _"W) are the coordinates (x _'W, y' before moving on a two-dimensional plane of the display device 4 _W) and on a straight line connecting the origin, before movement coordinates (x _'W, y _'W) will show only away position distance M _W from.

The movement of each word W displayed on the display device 4 will be described more specifically. As shown in FIG.
As a word W _I of item is "animal" is input, the display device 4, the "animal" is displayed as a word W _I item to the origin of the two-dimensional coordinates, a plurality of words are displayed around In this case, it is assumed that the word “dolphin”, which is an aquatic animal, has been moved away from the “animal” displayed at the origin by operating the operation device 5. In this case, the vector V _W of each word W displayed on the display device 4 is stored in the search target database 2.
It is represented as shown in FIG. Then, when the moved word “dolphin” is separated from the word “animal” of the item, the calculation unit 52 of the operation device 5 calculates the moving distance M _Sd of the word “dolphin”.

The moving distance M _Sd of the word “dolphin” is calculated from the coordinates of the word “dolphin” before and after the movement.
It is calculated based on equation (5). Then, _assuming that the vector of the moved word “dolphin” is V _Wd , the distance between the vector V _Wd and the vector V _W of each word W displayed on the display device 4 and the moving distance of the word “dolphin” The moving distance MW of each word _W is calculated from _MSd based on equation (6).

As described above, the word “dolphin” has the moving distance M
_When moved by _Sd, the moving distances of the word “rabbit” and the word “fur seal” displayed on the display device 4 are as follows. If the vector of the word "rabbit" is V _Wu ,
The vector V _Wu is (7,5,6,2,0,0) as shown in FIG.
Is represented by Also, when the vector of the word "seals" and V _Wt, the vector V _Wt is expressed in the (7,4,3,6,0,1) as shown in FIG. Further, _assuming that the vector of the word “dolphin” is V _Sd , the vector V
_Sd is represented by (7, 4, 2, 6, 1, 1) as shown in FIG. Then, the vector V _{Wd of} the moved word “dolphin”
The distance between the vector V _Wu of the word "rabbit" and dist (V _Wu, V
_Wd ) is calculated based on the above equation (1), and the distance dist ( _VWt , _VWd ) between the moved vector V _Wd of the word “dolphin” and the vector V _Wt of the word “fur _seal ” is calculated by the above ( 1) It is calculated based on the equation. The equations are shown in equations (8) and (9), respectively.

[0045]

Dist (V _Wu , V _Wd ) = ｛(7−7) ² + (5−4) ² + (6−2) ² + (2−6) ² + (0−1) ² + ( 0−1) ² ｝ ^1/2 = 5.92 …… (8)

[0046]

Dist (V _Wt , V _Wd ) = ｛(7−7) ² + (4−4) ² + (3−2) ² + (6−6) ² + (0−1) ² + ( 1-1) ² ｝ ^1/2 = 1.41 (9) In this way, the distance dist (V _Wu , V _Sd ) between the vector V _Wu of the word “rabbit” and the vector V _Sd of the word “dolphin”
And the distance dist (V _Wt , V _Sd ) between the vector V _Wt of the word “fur _seal ” and the vector V _Sd of the word “dolphin” are calculated, the calculation unit 52 calculates the moving distance of the word “dolphin”. Based on M _Sd , the moving distances M _Wu and M _Wt of the word “rabbit” and the word “fur seal” are calculated based on the above equation (5). Each arithmetic expression is (1
These are shown as equations (0) and (11), respectively. However, the moving distance M _Sd of the word “dolphin” is 4.0 and the coefficient α is 0.15.

[0047]

M _Wu = ｛1−0.15 · dist (V _Wu , V _Sd )｝ M _Sd = (1−0.15 · 5.92) · 4.0 = 0.45 (10)

[0048]

M _Wt = ｛1−0.15 · dist (V _Wt , V _Sd )｝ M _Sd = (1−0.15 · 1.41) · 4.0 = 3.15 (11) Then, the word “rabbit” and the word “fur seal” "But,
The coordinates of each of the words “rabbit” and “fur seal” after being moved on the two-dimensional plane of the display device 4 by the distances M _Wu and M _Wt , respectively, are calculated based on the above equation (6).

The word "rabbit" is displayed at the origin of the two-dimensional plane on the display device 4 because its vector V _Wu is far from the vector V _Sd of the word "dolphin" and the degree of similarity between the two is low. Item "animal"
Only a small travel distance of 0.45 can be kept away.
However, the word “fur seal” has a vector V _Wt that is close to the vector V _Sd of the word “dolphin” and has a high degree of similarity, so that the vector V _Wt is located at the origin of the two-dimensional plane displayed on the display device 4. The word "animal" in the displayed item is moved away by a relatively large distance of 3.15.

That is, when the word “dolphin” is moved away from the word “animal” of the item, the word of “fur seal”, which is the same aquatic creature as “dolphin,” is also changed from the word “animal” of the item. They can be separated by the same distance. In contrast, the land animal "rabbit"
Is only slightly farther away from the word "animal" in the item.

As described above, a word having a vector whose distance is closer to the vector of the word moved away from the word of the item by the operation of the operation device 5, in other words, similar to the word moved away from the word of the item. Words with a high degree of distance are greatly separated from each other as well as words that have been moved away. Conversely, words having a vector that is separated from the word of the item by the operation of the operation device 5 have a vector that is farther away from the word of the item. Then, a word having a low degree of similarity to a word distant from the item word has a smaller distance away from the item word. As a result, unnecessary words as search results are displayed on the display device 4 in a state where they are kept away from the word of the item, and the words required as search results can be efficiently narrowed down to the periphery of the item.

A word considered necessary as a search result is
When the object is moved so as to approach the coordinate origin, the operation is performed in the same manner as described above. In this case, the moving distance M _{S of the} word moved by the operation device 5 is displayed on the display device 4. the moving distance M _W of each word has become a negative value.

Therefore, when the word “dolphin” is moved to approach the coordinate origin by operating the operation device 5, the word “fur seal” having a high degree of similarity to the word “dolphin” greatly approaches the coordinate origin. Conversely, the word "rabbit", which has a low degree of similarity to the word "dolphin", only very slightly approaches the coordinate origin.

With respect to the display state of the display device 4 shown in FIG. 4, assuming that a small animal on land is necessary as a search result, as shown in FIG. 9, the words "dolphin" and "lion" are added. The word “squirrel” is moved in a direction approaching the coordinate origin while moving away from the coordinate origin. Then, each word displayed on the display device 4 is:
Each was moved as indicated by the arrow or kept stationary. As a result, the display by the display device 4 changed as shown in FIG. From FIG. 10, the words “squirrel” and “rabbit”, which are small animals on land, are in proximity to the word “animal” displayed at the coordinate origin, and these are necessary as search results. It is easily recognized that the words are narrowed down.

In the above embodiment, the word as each data in the database is represented by a vector in which the relative similarity is quantified for each of the six elements, and the degree of similarity of each data is determined by Although the determination is made based on the distance between the vectors, the present invention is not limited to such an embodiment. For example, a plurality of keywords are set for each data, and similarity is determined based on the degree of matching of the keywords between the data. May be determined. In the above embodiment, the search result is displayed on the display device in two-dimensional coordinates. However, the present invention is not limited to such a display method. Although the moving device is configured to be moved by an operation device configured by a pointing device such as the above, the present invention is not limited to such a structure.

[0056]

According to the data retrieval apparatus of the present invention, when a part of the data displayed as a retrieval result by the display means is moved by the operation means, the degree of similarity to the moved data is changed. Since high data is also moved in the same direction, necessary data as a search result can be narrowed down so as to be displayed in a state close to the search target item, thereby significantly improving search efficiency.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an example of a data search device of the present invention.

FIG. 2 is a flowchart illustrating the operation of the data search device.

FIG. 3 is a table showing vectors of data stored in a search target database of the data search device.

FIG. 4 is a schematic diagram showing an example of a display screen of a display device of the data search device.

FIG. 5 is a table showing an example of data stored in a buffer of the data search device.

FIG. 6 is a block diagram showing details of an operation device of the data search device.

FIG. 7 is a flowchart for explaining the operation of the operation device of the data search device.

FIG. 8 is a table showing vectors stored in a search target database for data displayed on a display device of the data search device.

9 is a schematic diagram for explaining movement of data on a display screen of the display device shown in FIG.

FIG. 10 is a schematic diagram showing a result of moving each data on the display screen of the display device shown in FIG. 9;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Input device 2 Search target database 3 Search device 4 Display device 5 Operation device 51 Input part 52 Operation part 53 Output part

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G06F 17/30

Claims (3)

(57) [Claims] 1. A search target database storing a plurality of data sets each having a relative similarity, input means for inputting a search request item, and each data stored in the search target database. Determining the degree of similarity with the search request item input by the input unit based on the similarity of each data and the search request item to extract a plurality of data similar to the search request item Search means; display means for displaying a plurality of data extracted by the search means together with the search request item; and data displayed on the display means so as to approach or move away from the search request item. Operation means for moving the data to
The degree of similarity with other data displayed on the display means is determined by the similarity of each data, and based on the determination, the other data displayed on the display means is compared with the search target item. A data search device characterized by being moved. 2. The data search device according to claim 1, wherein the data displayed on the display means has a longer moving distance as the degree of similarity to the data moved by the operation means is higher. 3. The data search device according to claim 1, wherein each of the data stored in the search target database is represented by a vector in which the similarity is quantified for each of a plurality of elements.